Post-selected Criticality in Measurement-induced Phase Transitions

This paper demonstrates that explicit post-selection in measurement-induced phase transitions fundamentally alters the universality class—yielding a distinct correlation-length exponent and negative effective central charge—by reweighting rare trajectories, while also establishing an equivalence with Random Tensor Network entanglement transitions and identifying a minimum onsite dimension of three as a necessary condition for the transition.

The Gist

Imagine you are watching a complex magic trick performed by a team of quantum "magicians" (particles). Usually, these magicians work together in perfect harmony, creating a massive, tangled web of connections (entanglement) that spreads across the whole room. This is the Volume Law phase: information is everywhere, and it's incredibly hard to figure out what any single magician is doing just by looking at them.

However, imagine a critic (a "measurement") starts peeking at the magicians. If the critic peeks too often, the magic breaks. The magicians stop cooperating, the web untangles, and the room falls into silence. This is the Area Law phase: information is trapped locally, and the magic is gone.

The point where the magic switches from "tangled" to "untangled" is called a Measurement-Induced Phase Transition (MIPT).

The Problem: The "Unlucky" Critic

In the real world, when a critic peeks, they see a random outcome. Sometimes they see "Heads," sometimes "Tails." To understand the transition, scientists usually have to average out all possible outcomes. But here's the catch: the most interesting, dramatic moments of the transition happen in very specific, rare sequences of outcomes. Finding these sequences is like trying to find a specific grain of sand on a beach by looking at the whole beach at once. It's nearly impossible.

The Solution: "Post-Selection" (The Editor's Cut)

This paper introduces a clever trick called Post-Selection. Instead of watching the random show, the scientists say, "Let's only watch the versions of the magic trick where the critic always sees 'Heads'."

Think of it like editing a movie. In a standard movie, you see every scene. In this "Post-Selected" version, the editor cuts out every scene where the critic sees "Tails" and only keeps the "Heads" scenes. This forces the system into a specific, rare path.

What They Discovered

By forcing the system to follow this specific "Heads" path, the scientists found something surprising: The rules of the game changed completely.

1. A New Kind of Criticality: The transition didn't just get slightly different; it became a totally new type of physics. It's like discovering that if you only watch movies where the hero wins, the story follows a completely different set of laws than if you watch all movies.
2. The "Negative" Heart: In physics, there's a number called the "central charge" that measures how complex a system is. Usually, this number is positive. But in this post-selected world, they found a negative number.
   • Analogy: Imagine a scale that usually weighs things positively. In this new world, the scale says the "weight" of the complexity is negative. It's a bizarre, counter-intuitive result that suggests this new phase of matter is fundamentally different from anything we've seen before.
3. Two Different Roads, Same Destination: The scientists compared this "Post-Selected" quantum circuit to a mathematical model called a Random Tensor Network (think of it as a giant, 3D puzzle made of random blocks). Surprisingly, even though one model looks like a movie script and the other like a 3D puzzle, they both fell into the exact same "universality class." They are different languages describing the same underlying reality.

The "Qubit vs. Qutrit" Mystery

The paper also tested a simpler version where the rules are perfectly repeating (no randomness in the timing). In the random versions described above, qubits work fine. But in this non-random version:

• The Qubit (2-state) Failure: When they used standard 2-state particles (qubits, like a coin that is Heads or Tails), the magic trick failed to transition. It just stayed in one state.
• The Qutrit (3-state) Success: When they upgraded to 3-state particles (qutrits, like a coin that can be Heads, Tails, or Standing on its edge), the transition happened again!
  • Analogy: It's like trying to balance a pencil on its tip. A 2-sided coin can't balance in a third state, so it just falls over. But a 3-sided object has just enough complexity to find a stable, critical balance point.

Why Does This Matter?

This research is a big deal for two reasons:

1. It solves a practical problem: By using "forced" measurements (post-selection), we can study these quantum transitions without needing to wait for impossibly rare random events. It's like using a filter to see the signal clearly without the noise.
2. It reveals new physics: The fact that the transition changes its "personality" (universality class) when we force a specific outcome tells us that the "rare" paths in quantum mechanics aren't just noise—they hold the key to entirely new types of matter and information storage.

In short: The paper shows that if you force a quantum system to follow a specific, rare path, it doesn't just behave "better" or "worse"—it transforms into a completely new species of physics with its own unique laws, negative complexities, and—in the absence of randomness—a need for slightly more complex building blocks (3-state particles) to function.

Technical Summary

1. Problem Statement

Measurement-induced phase transitions (MIPTs) describe the competition between unitary dynamics (which generates entanglement) and local measurements (which suppress it). Standard MIPTs are defined by averaging over measurement outcomes weighted by the Born rule. However, detecting these transitions experimentally or numerically is challenging because the transition is a property of individual quantum trajectories, requiring non-linear observables that are exponentially rare to sample.

The paper addresses the nature of MIPTs under forced post-selection, where the system is conditioned on a specific, fixed measurement outcome (e.g., always projecting to $|0\rangle$) rather than sampling according to Born probabilities. The central questions are:

• Does explicit post-selection alter the universality class of the transition?
• How does this "post-selected" MIPT relate to the entanglement transitions observed in Random Tensor Networks (RTNs)?
• Is spatio-temporal randomness essential for the criticality, or can it persist in translationally invariant, time-periodic circuits?

2. Methodology

The authors employ numerical simulations and statistical mechanics mappings to compare two distinct models:

A. Post-Selected Random Quantum Circuits (P-RQC)

• Setup: A 1D chain of qubits evolving under a brick-layer architecture of random two-qubit unitary gates (drawn from the Haar distribution).
• Dynamics: At each time step, qubits are subjected to "forced" measurements with probability $p$. Unlike standard MIPTs, these measurements are not sampled probabilistically; they are deterministically projected onto the $|0\rangle$ state.
• State Evolution: The state evolves as $|\psi_{t+1}\rangle = \prod M_i \prod U_{i,i+1} |\psi_t\rangle$, where $M_i$ is the projection operator. The randomness arises solely from the unitary gates and the random placement of measurement locations, not from the measurement outcomes.

B. Random Tensor Networks (RTN)

• Setup: A 2D network of local tensors with Gaussian-distributed entries.
• Mapping: The authors map the P-RQC to an RTN by tuning the effective bond dimension ($D_{eff}$) of the network legs. This is controlled by the second Rényi mutual information of the edge states.
• Goal: To demonstrate that the RTN entanglement transition (driven by bond dimension) and the P-RQC transition (driven by measurement rate) belong to the same universality class.

C. Translationally Invariant Variants

• To test the necessity of randomness, the authors construct time-periodic, translationally invariant circuits where a single unitary gate (or tensor) is repeated throughout the spacetime lattice. They test both qubits (2-level) and qutrits (3-level) systems.

D. Diagnostic Tools

The authors utilize several critical observables to locate the phase transition and extract critical exponents:

• Tripartite Mutual Information ($I_3$): Used to distinguish between area-law (disentangled) and volume-law (entangled) phases.
• Ancilla Entropy ($S_a$): An ancilla qubit is entangled with the system to probe purification dynamics and local order parameters.
• Finite-Size Scaling: Analysis of scaling relations (e.g., $I_3 \sim f(L^{1/\nu}(p-p_c))$) to extract critical points ($p_c$) and exponents ($\nu, \beta, \eta, z$).
• Effective Central Charge ($c_{eff}$): Derived from the finite-size scaling of the quenched free energy density, $F(L) \sim f(\infty) - \frac{\pi c_{eff}}{6L^2}$, using the replica limit $k \to 0$.

3. Key Contributions

1. Identification of a New Universality Class: The paper establishes that post-selected MIPTs belong to a distinct universality class from standard Born-rule MIPTs. The forced reweighting of trajectories fundamentally changes the critical behavior.
2. Equivalence with RTNs: The authors demonstrate a rigorous mapping showing that post-selected MIPTs and RTN entanglement transitions share the same critical exponents and universality class, validating the statistical mechanics framework where both map to a symmetry-breaking transition in a replica model.
3. Discovery of Negative Effective Central Charge: A major theoretical finding is the calculation of a negative effective central charge ($c_{eff} \approx -0.4$), which is highly unusual for disordered systems and implies specific properties of the underlying conformal field theory (CFT).
4. Dimensionality Dependence: The study reveals that for translationally invariant (non-random) circuits, a transition only occurs for qutrits (onsite dimension $D \geq 3$), not qubits ($D=2$). Importantly, this dimensional constraint applies only to the translationally invariant case; in the random circuit, qubits are sufficient. This suggests that while randomness aids criticality, it is not strictly necessary if the local Hilbert space dimension is sufficiently large.

4. Key Results

Critical Exponents and Parameters

The authors extracted the following critical exponents for the Post-Selected RQC (P-RQC) and RTN, comparing them to the standard Born-rule RQC (B-RQC):

Parameter	P-RQC (Post-Selected)	RTN	B-RQC (Born-Rule)	Percolation
Critical Point ($p_c$ or $\chi_c$)	$0.24(1)$	$0.9(9)$	$0.168(5)$	$0.5$
Correlation Length ($\nu$)	$2.1(5)$	$2.2(5)$	$1.2(2)$	$1.33$
Dynamical Exponent ($z$)	$0.96(5)$	$1.05(5)$	$1.06(4)$	$1$
Order Parameter ($\beta$)	$0.16(5)$	$0.16(2)$	$0.10(4)$	$0.139$
Anomalous Dimension ($\eta$)	$0.17(2)$	$0.16(2)$	$0.19(1)$	$0.21$
Effective Central Charge ($c_{eff}$)	$-0.40(6)$	$-0.35(2)$	$0.25(3)$	N/A

• Universality: The exponents for P-RQC and RTN match within error bars, confirming they are in the same universality class.
• Stability: The correlation length exponent $\nu \approx 2.1$ is greater than 2. According to Harris' criterion, this implies the transition is stable against static disorder, unlike the Born-rule MIPT which flows to infinite randomness.
• Negative $c_{eff}$: The negative value arises because the replica partition function $Z_k$ has trivial limits at $k=0$ and $k=1$, and the function $c(k)$ is continuous without additional zeroes, forcing the slope at $k=0$ to be negative.

Translationally Invariant Models

• Qubits: No phase transition was observed in the translationally invariant qubit circuit; the system remained in the volume-law phase or showed no critical scaling.
• Qutrits: A clear phase transition was observed in the translationally invariant qutrit circuit. This indicates that onsite Hilbert space dimension is a crucial factor for MIPT criticality in the absence of spatio-temporal randomness.

5. Significance

• Theoretical Insight: The work resolves the "pathological" view of post-selected MIPTs, reframing them as a natural, distinct universality class of information-theoretic transitions. It bridges the gap between quantum circuit dynamics and tensor network physics.
• Experimental Relevance: While standard MIPTs are hard to detect due to the exponential rarity of specific trajectories, post-selected models (or forced measurements) offer a theoretical framework where the transition is robust and potentially more accessible to specific experimental protocols (e.g., using cross-entropy benchmarking or specific post-selection strategies).
• New Physics: The discovery of a negative effective central charge challenges conventional wisdom about disordered systems and suggests new types of non-unitary conformal field theories.
• Robustness: The finding that $\nu > 2$ suggests these post-selected phases are robust against static disorder, a property not shared by standard Born-rule MIPTs.

In conclusion, the paper demonstrates that by altering the sampling procedure from Born-rule to post-selection, one accesses a fundamentally different critical regime characterized by negative central charge, stability against disorder, and a dependence on local Hilbert space dimension that persists even in the absence of randomness.